Crane inspection methods, devices, and electronic equipment based on artificial intelligence

Abstract

This application provides an artificial intelligence-based crane inspection method, device, and electronic equipment, relating to the field of crane inspection. By introducing a crane intelligent agent cluster based on a large model, it achieves intelligent verification and closed-loop control of the entire process, including inspection task configuration, path generation, data acquisition and verification, multi-source data collaborative processing, and equipment status diagnosis. The intelligent agent cluster can perform dual verification of inspection tasks and collected data, ensuring the safety and validity of the inspection process. Combined with the deployment model, it automatically generates optimized inspection paths and utilizes graph neural networks to fuse multi-source features and knowledge networks, significantly improving the accuracy and robustness of equipment status identification. Simultaneously, it predicts the remaining lifespan and failure probability of components through historical data analysis and maps the structured inspection results to a virtual model in real time to generate a visual report, greatly improving the intelligence level, operation and maintenance efficiency, and decision support capabilities of crane inspection.

Description

Technical Field

[0001] This application relates to the field of crane inspection, and more specifically, to a crane inspection method, device, and electronic equipment based on artificial intelligence. Background Technology

[0002] With the rapid development of the construction industry, lifting machinery and equipment such as tower cranes and hoists play a crucial role in various engineering projects. However, traditional lifting machinery management methods have many shortcomings, including but not limited to unclear asset information, non-standard management processes, and difficulty in real-time monitoring of equipment status, resulting in low asset management efficiency and increased operating costs for enterprises. Summary of the Invention

[0003] The purpose of this application is to provide an artificial intelligence-based crane inspection method, device, and electronic equipment, aiming to achieve intelligent, automated, and visualized precise inspection and result push of crane equipment components throughout the entire process.

[0004] Firstly, an artificial intelligence-based crane inspection method is provided, which may include: Obtain the inspection task of at least one equipment component of the target crane and input each inspection task into a crane agent cluster created based on a large model; The crane intelligent agent cluster verifies the inspection task. After successful verification, it calls the deployment model corresponding to the target crane to generate an inspection path, triggering the inspection equipment to collect component data at each inspection point according to the inspection path. The crane intelligent agent cluster then verifies the collected component data. The inspection path includes the inspection points corresponding to each inspection task. The deployment model is a virtual model of the target crane. When the component data meets the task conditions of the inspection task and passes the verification of the crane intelligent agent cluster, the crane intelligent agent cluster performs collaborative processing on the component data to determine the equipment status information of the corresponding inspection point. Based on the equipment status information, a structured inspection result is generated, the inspection result is mapped to the location of the corresponding inspection point in the deployment model, and the inspection results of multiple inspection points are summarized to generate a visual inspection report and pushed to the user terminal.

[0005] In one possible implementation, the path planning agent in the crane agent cluster determines the equipment location of the inspection equipment as the starting point of the inspection path. Based on the location of each inspection point in the deployment model and the starting point, and combined with the verification results of each agent in the crane agent cluster on obstacle information, safety constraints and historical path efficiency, the inspection sequence of each inspection point is determined. The inspection path is determined based on the inspection sequence and the location of each inspection point.

[0006] In one possible implementation, the inspection sequence of each inspection point is determined based on its location in the deployment model and the location of the equipment, including: The coordinates of any two inspection points are processed to determine the straight-line optical path data of the two inspection points in the deployment model; the coordinates of the inspection points are the coordinates of the inspection points in the equipment coordinate system of the inspection equipment. A ray tracing algorithm is used to process the straight-line optical path data and the layout model. If the straight-line optical path data does not intersect with the obstacles in the layout model, then the corresponding two inspection points are determined as adjacent inspection point pairs. The safety verification agent in the crane agent cluster performs a safety assessment on each candidate adjacent inspection point pair, and the path planning agent determines the inspection sequence of each inspection point based on the distance between the starting point of the inspection path and each inspection point in the inspection point pair, the constraints of each inspection point, and the safety assessment results.

[0007] In one possible implementation, the process of determining the inspection task includes: Receive task configuration information input by the user, the task configuration information including image acquisition conditions and image location conditions; The task parsing agent in the crane agent cluster performs semantic understanding and consistency verification on the task configuration information to determine the inspection task.

[0008] In one possible implementation, the task conditions include inspection location conditions and data conditions; The component data includes multi-source data to be processed and the corresponding acquisition locations; The crane intelligent agent cluster performs real-time verification of the collected component data, including: the data verification intelligent agent in the crane intelligent agent cluster determines whether the collection location of the multi-source data meets the inspection location conditions, and whether the multi-source data meets the data conditions. If both conditions are met, then the component data satisfies the task conditions of the inspection task and passes verification.

[0009] In one possible implementation, the multi-source data includes image data of the target crane's components in different spectral bands, as well as temperature data, ultrasonic data, and point cloud data of the target crane's components. The crane intelligent agent cluster collaboratively processes the component data to determine the equipment status information at the corresponding inspection points, including: The feature extraction agent in the crane intelligent agent cluster extracts features from image data collected at the same time under different spectral bands, as well as temperature data, ultrasonic data and point cloud data of the target crane components, and determines the corresponding feature vectors. The fusion agent fuses the feature vectors to obtain a multidimensional feature vector; A knowledge network is constructed using each equipment component as a node and the connection relationships between each equipment component, historical fault records, and maintenance records as edges. The state diagnosis agent in the crane agent cluster uses a trained graph neural network to process the multidimensional feature vector and the knowledge network, and after confirmation by the cluster consensus mechanism, determines the equipment state information.

[0010] In one possible implementation, after determining the device status information, the method further includes: For any inspection point, the predictive agent in the crane agent cluster obtains multiple historical component data of that inspection point within a historical time window. Analyze multiple historical component data to determine the changing trend of the equipment components corresponding to the inspection point; Based on the aforementioned trends, one or more of the following are predicted: the remaining service life of the equipment components, the probability of failure, and the performance degradation trend. The prediction results are then added to the structured inspection results after being verified by the cluster.

[0011] Secondly, an artificial intelligence-based crane inspection device is provided, the device comprising: The acquisition module is used to acquire the inspection tasks of at least one equipment component of the target crane and input each inspection task into the crane agent cluster created based on the large model. The verification module is used to verify the inspection task by the crane intelligent agent cluster. After the verification is successful, the module calls the deployment model corresponding to the target crane to generate an inspection path, triggers the inspection equipment to collect component data at each inspection point according to the inspection path, and the crane intelligent agent cluster verifies the collected component data. The inspection path includes the inspection points corresponding to each inspection task. The deployment model is a virtual model of the target crane. The collaboration module is used to determine the equipment status information of the corresponding inspection point by having the crane intelligent agent cluster collaboratively process the component data when the component data meets the task conditions of the inspection task and passes the verification of the crane intelligent agent cluster. The mapping module is used to generate structured inspection results based on the equipment status information, map the inspection results to the corresponding inspection points in the deployment model, and summarize the inspection results of multiple inspection points to generate a visual inspection report and push it to the user terminal.

[0012] Thirdly, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.

[0013] Fourthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of any of the methods described in the first aspect above.

[0014] This application provides an artificial intelligence-based crane inspection method. This method introduces a cluster of crane intelligent agents created based on a large model, achieving intelligent verification and closed-loop control throughout the entire process, including inspection task configuration, path generation, data acquisition and verification, multi-source data collaborative processing, and equipment status diagnosis. The intelligent agent cluster can perform dual verification of inspection tasks and collected data, ensuring the safety and validity of the inspection process. Combined with the deployment model, it automatically generates optimized inspection paths and utilizes graph neural networks to fuse multi-source features and knowledge networks, significantly improving the accuracy and robustness of equipment status identification. Simultaneously, it predicts the remaining lifespan and failure probability of components through historical data analysis and maps structured inspection results to a virtual model in real time, generating a visual report, greatly improving the intelligence level, operational efficiency, and decision support capabilities of crane inspection. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A system architecture diagram for an artificial intelligence-based crane inspection method provided in this application embodiment; Figure 2 This is a schematic diagram of the interface for joining an organization provided in an embodiment of this application; Figure 3A schematic diagram of the interface for creating a new enterprise / organization / team provided in the embodiments of this application; Figure 4 A schematic diagram of the interface of the personnel management module provided in an embodiment of this application; Figure 5 This is a schematic diagram of the interface for the role management function provided in an embodiment of this application; Figure 6 A schematic diagram of the interface for configuring permissions for adding roles provided in an embodiment of this application; Figure 7 A schematic diagram of the interface of the construction crane inspection platform provided in this application embodiment; Figure 8 A schematic diagram of the interface for creating a task provided in an embodiment of this application; Figure 9 A schematic diagram of the interface for providing basic task information in an embodiment of this application; Figure 10 A schematic diagram of the interface for providing basic node information in an embodiment of this application; Figure 11 A schematic diagram of the task editing interface provided in this application embodiment; Figure 12 A schematic diagram of the interface for auditing inspection points provided in this embodiment of the application; Figure 13 A flowchart illustrating an artificial intelligence-based crane inspection method provided in this application embodiment; Figure 14 A schematic diagram of the structure of an artificial intelligence-based crane inspection device provided in an embodiment of this application; Figure 15 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0018] The crane inspection method based on artificial intelligence provided in this application embodiment can be applied to... Figure 1 In the system architecture shown, such as Figure 1 As shown, the system may include: a server and an intelligent inspection system for lifting machinery.

[0019] The intelligent inspection system for lifting machinery includes: a login module, a personnel management module, a construction crane asset management platform, and a construction crane inspection platform. The construction crane asset management platform may include procurement management, transportation management, and installation and commissioning modules; the construction crane inspection platform may include task modules, inspection modules, and data processing modules.

[0020] Existing users can log in to the intelligent inspection system for lifting machinery by entering their account and password in the login interface of the login module. New users can create organizations / companies / teams or join visitor organizations through the registration interface in the login module, or they can join the corresponding organization through an organization invitation code.

[0021] Combination Figure 2 As shown, A. The process of entering the visitor organization is as follows: After a new user logs into the system, a dialog box for creating an organization / company / team pops up, triggering the "Enter Visitor Organization" button. The new user enters the visitor organization and can browse functions within the visitor organization.

[0022] B. After a new user logs into the system, a dialog box for creating an organization / company / team will pop up. After obtaining an organization invitation code from the corresponding organization administrator, the user will enter the organization invitation code in the organization invitation code input box, trigger the "Join Organization" button, and join the invited organization. At the same time, the user will log into the system.

[0023] C. Combination Figure 3 As shown, the process of creating an organization / company / team is as follows: After a new user logs into the system, a dialog box for creating an organization / company / team pops up. Triggering the "Create Organization" button will bring up another dialog box for creating an organization / company / team. In the dialog box, enter the organization name, select the organization type, organization size, and organization location, and click the "Agree to the Agreement and Register" button to create a new organization / company / team and log in to the system at the same time.

[0024] The intelligent inspection system for lifting machinery is an integrated information platform for the full lifecycle management of tower cranes and construction hoists. The system sets up corresponding functional modules around each key work node from equipment procurement and warehousing, installation and commissioning, operation, regular inspection, maintenance, to scrapping, achieving comprehensive, standardized, and traceable asset management and operation and maintenance support.

[0025] This system incorporates a role-based access control (RBAC) mechanism tailored to the organizational structure of tower crane companies, adapting to the division of responsibilities among personnel in different departments at different business stages. System administrators can set up corresponding user roles based on actual positions and configure functional permissions for each role corresponding to its responsibilities.

[0026] For example, the transportation department only participates in transportation-related processes such as equipment allocation and transfer. Therefore, its role and permissions are limited to the corresponding functional module of "transportation management" and it can only perform operations related to transportation tasks. On the other hand, the inspection and maintenance department is mainly responsible for equipment status inspection, fault reporting and maintenance records. Its permissions are concentrated in functions such as "inspection management" and "maintenance management".

[0027] Furthermore, the system administrator has the highest level of authority, possessing full permissions over all modules of the system, including the ability to view, create, edit, modify, and delete each module.

[0028] Combination Figure 4 As shown, the personnel management module provides system administrators with comprehensive tools to manage users of the system and flexibly assign roles and permissions according to organizational needs.

[0029] System administrators can quickly copy a unique company invitation link by clicking the "Generate Organization Invitation Link" button. Any new user who receives this link and completes registration using it will be automatically added to the company's personnel management system and become a full member.

[0030] By entering keywords (such as username, email address, etc.) in the "Search" field, the system can filter and display a list of matching users in real time, making it easy to quickly locate a specific user.

[0031] Administrators can select specific role tags as filters using the "Role" option to display only all users with that role. This method facilitates quick viewing and management of user groups with the same responsibilities or permission levels.

[0032] The "Status" item allows administrators to filter users' activation status (e.g., enabled, disabled) to identify which accounts are active and which require further processing (e.g., reactivation or deletion).

[0033] Disable the corresponding user from using the system by triggering the "Disable" button in the "Action" section. Modify the user's role type by triggering the "Modify" button.

[0034] Combination Figure 5 As shown, the system can also include a role management function, which is used to configure and manage company roles. For example, the entire asset management process includes operation management, allocation management, transportation management, installation and dismantling management, inspection management, etc. Roles can be added in the above management functions, and the corresponding permissions can be set for each role.

[0035] Within the system, administrators can flexibly create and adjust user roles through the role management function, and assign corresponding functional permissions to each role to achieve granular access control.

[0036] When an administrator triggers the "Add Role" button, a new custom role can be created. Combined with... Figure 6 As shown, during the creation process, the system provides a permission configuration interface for adding roles. Administrators can select the functional modules that a role can access based on actual needs. The selected functional modules indicate that the role has the corresponding operation permissions. For existing roles, administrators can access the permission modification interface by triggering the "Edit" button in the "Operations" column, and readjust the role's access and operation permissions for each functional module. This function supports dynamic updates of permissions, ensuring that the system can flexibly adapt to changes in organizational structure and business processes.

[0037] Combination Figure 7 As shown, the construction crane inspection platform provides unified lifecycle management functions for crane inspection tasks in all projects, covering the entire process from task creation to final archiving.

[0038] Triggering the "Create Task" option will bring up the task creation page, combined with... Figure 8 As shown, select a project, enter the task name, select the tower crane to be inspected in the "Tower Crane Number" field, check the corresponding personnel and equipment in the "Inspection Personnel" or "Inspection Equipment" field, select the inspection start time and inspection end time from the system, and then create an inspection task in the project. The inspection task will then enter the execution module.

[0039] Inspection tasks are divided into multiple status nodes based on their execution progress and processing status: Nodes to be released: Inspection tasks that have been created but have not yet been officially issued for execution; In-process node: Inspection tasks that are first created and published enter this state by default; combined with Figure 9 As shown, the "View" option in this state allows you to view the details of this inspection task. The inspection points for this task include: foundation, standard section, attachment, slewing mechanism, counterweight boom, hoisting mechanism, luffing mechanism, trolley, boom, and hook. Each inspection task has an inspection start time and an inspection end time. Managers and inspection personnel can view the inspection details in the "Details" option.

[0040] Furthermore, in combination Figure 10 As shown, by triggering the "Details" option, you can view the specific details of the inspection task.

[0041] By triggering Figure 7 The "Edit" option in the text, combined with Figure 11 As shown, the task name, inspection personnel or equipment, and inspection end time of the inspection can be edited.

[0042] "Under Review" stage: After the inspection personnel complete and submit the inspection results via the mini-program or client, or after the inspection results are automatically generated, the inspection task automatically moves to the "Under Review" stage; combined with Figure 12 As shown, the auditors reviewed each inspection point in the inspection task one by one.

[0043] Trigger the "Review" button in the details section of each inspection point to bring up the inspection results. If there are no abnormalities in the inspection results details, trigger the "Submit for Review" option in the review section. If there are abnormalities at the inspection point that need to be addressed, trigger the "Abnormalities" option in the review section, fill in the comments that require rectification, check the "Rectify" button, and select the rectification deadline. After reviewing all inspection points, click the "Submit for Review" button.

[0044] Inspection tasks that have been fully reviewed and found to be without any abnormalities are considered normal and will be moved to the historical task list. Inspection tasks with abnormalities at inspection points that require handling will be moved to the rectification stage.

[0045] In Progress of Rectification: If rectification is required during the review process, the system can mark the relevant items as rectification items. Inspection tasks containing rectification items will automatically enter the "In Progress of Rectification" status, awaiting completion by the responsible personnel. During rectification, the system will display the inspection tasks requiring rectification. Clicking the "View" button will show details of the specific inspection points requiring rectification. Rectification-required inspection tasks are also pushed to the mini-program or client. After on-site processing, the results can be resubmitted for review on the mini-program or client.

[0046] Historical Tasks: When all rectification items for any inspection task have been completed, or when it has been reviewed and confirmed that no rectification is required, the task will be automatically archived to the "Historical Tasks" module for subsequent querying and tracing.

[0047] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0048] Figure 13 This is a flowchart illustrating an artificial intelligence-based crane inspection method provided in an embodiment of this application.

[0049] This solution is a method for automating crane inspection by building a multi-agent collaborative cluster based on a large-scale artificial intelligence model. The core is to complete the entire process of task verification, path planning, data processing, and status diagnosis through the division of labor and collaboration among the large-scale model's intelligent agents, ultimately outputting visualized inspection results. All core terms and their corresponding steps are explained below: Large model: refers to a pre-trained artificial intelligence model with a large parameter scale and strong semantic understanding and logical reasoning capabilities. In this solution, it is used to build intelligent agents that can divide tasks and cooperate. Crane agent cluster: A collaborative network of multiple independent agents responsible for different inspection functions, created based on a large model. Each agent is responsible for a specific task and together they complete the entire inspection process. Inspection task: For a specific equipment component of the crane that needs to be inspected (such as wire rope, drum, slewing bearing, brake, etc.), clarify what needs to be inspected and what data needs to be collected.

[0050] like Figure 13 As shown, the method may include: S1310, obtain the inspection task and input it into the crane intelligent agent cluster.

[0051] First, collect the inspection tasks for all equipment components of the target crane that need to be inspected. Then, input all tasks in batches into a pre-built crane intelligent agent cluster based on a large model, and the cluster will initiate subsequent processing.

[0052] For example, if the target crane is a quay container crane, it is necessary to check three items: wire rope wear, brake clearance, and slewing bearing cracks. The system will organize these three inspection tasks and input them into a crane agent cluster composed of six different agents, such as a task parsing agent and a path planning agent.

[0053] S1320, after the cluster verification task, calls the deployment model to generate the inspection path, triggering the collection and verification of component data.

[0054] Deployment Model: A digital virtual twin model of the target crane, which fully restores the size, position and structure of all crane components, includes the coordinate information of all pre-set inspection points, and can also carry subsequent inspection result mapping to achieve visualization; Inspection path: The inspection equipment starts from the starting point and travels through all the inspection points that need to be checked in sequence, including all the inspection points corresponding to the tasks. Inspection points: Pre-marked locations on the deployment model where data collection for the corresponding components is required; Inspection equipment: Mobile inspection robots equipped with image acquisition sensors, temperature sensors, ultrasonic sensors, and lidar (generally installed on movable tracks on the main beam of a crane, capable of autonomous movement); Component data: Various test data of the corresponding equipment components collected by the inspection equipment at each inspection point.

[0055] The crane intelligent agent cluster first verifies the format and logical rationality of the input inspection task. After the verification is passed, it calls the virtual deployment model of the target crane, generates a qualified inspection path based on the model information, and then issues instructions to the inspection equipment to move to each inspection point according to the inspection path and collect data from the corresponding components. After the data collection is completed, the intelligent agent cluster verifies the compliance of the collected data.

[0056] For example, after the intelligent agent cluster verifies that the logic of the three inspection tasks is correct, it calls the 1:1 three-dimensional twin deployment model of the quay bridge. The model has marked the wire rope wear inspection points (coordinates x1y1z1), the brake clearance inspection points (x2y2z2), and the slewing bearing crack inspection points (x3y3z3). Based on these points, the cluster generates an inspection path starting from the current standby position of the inspection robot. The robot sets off and collects data at the three points in sequence. After the data collection is completed, the cluster verifies whether the data is qualified.

[0057] S1330: After data verification is successful, the cluster collaboratively processes and determines the device status information.

[0058] Task conditions: The inspection task has requirements for data collection location and data quality. Only data that meets the requirements can be used for status diagnosis. Equipment status information: The health status of the equipment components corresponding to each inspection point is generally divided into categories such as normal, slight wear, crack defects, abnormal overheating, and fault warning, and also includes specific information such as the size and severity of the defects.

[0059] Once the collected component data meets the requirements of the inspection task and is verified by the intelligent agent cluster, the intelligent agents in the cluster, each responsible for different processing links, work together to process the data and ultimately determine the equipment status information of the component corresponding to each inspection point.

[0060] For example, the data collected by the robot at the three points all meet the location and quality requirements. The feature extraction agent, fusion agent, and state diagnosis agent in the cluster process the data in sequence, and finally obtain the following results: the wire rope wear is 1mm, which is normal; the brake clearance is 2mm, which meets the standard and is normal; the slewing bearing has a surface crack with a length of 5mm, which is abnormal and needs attention.

[0061] S1340 generates structured inspection results, maps them to the deployment model, and pushes a visual report.

[0062] Structured inspection results: The equipment status information obtained from the inspection is organized into a unified format, including standardized fields such as inspection points, component names, status classification, and defect parameters, which facilitates storage and display. Visualized inspection report: The inspection results are combined with the virtual deployment model to present them in a visual way. Users can intuitively see where there is a problem and the severity of the problem, which is clearer than a plain text report.

[0063] The system compiles equipment status information from all inspection points to generate standardized, structured results. The results for each point are then placed into the corresponding location in the deployment model. Finally, all results are summarized to generate a visual inspection report, which is then pushed to users' mobile phones, computers, and other terminals.

[0064] For example, the status information of the three points can be organized into a structured table, and the abnormal status of "5mm crack in slewing bearing" can be marked at the corresponding x3y3z3 position in the layout model. Users can open the WPS report or 3D visualization page on their mobile phones, click on the abnormal marker to view detailed parameters, and the complete report will be automatically pushed to the mobile terminal of the person in charge of equipment management.

[0065] In some embodiments, the steps for generating inspection paths are described in detail below: The path planning agent determines the starting point of the inspection path. The path planning agent sets the current actual location of the inspection equipment as the starting point of the entire inspection path.

[0066] Path planning agent: In the crane agent cluster, a specialized agent is responsible for calculating and generating the optimal inspection path.

[0067] The order of inspection points is determined by combining multiple verification results. Based on the positions of all inspection points in the deployment model and the determined starting point, the path planning agent hands over the initially generated multiple path candidate schemes to the corresponding agents in the cluster to verify obstacles, safety constraints, and path efficiency. Finally, the order of inspection of all inspection points is determined by combining the verification results.

[0068] Obstacle information: Structures on the crane recorded in the deployment model that are not inspection components and would obstruct the movement of the inspection robot or data collection, such as the crane's crossbeams and supporting angle steel; Safety constraints: The inspection path must meet safety requirements, such as not entering the high-risk operation area of ​​the crane, not colliding with moving parts, and the path must meet the slope restrictions of the inspection robot. Historical path efficiency: Efficiency data such as travel time and data collection time of different path schemes in similar inspection tasks in the past are used to optimize the current path.

[0069] The final inspection path is generated based on the sequence and location of the points. Following the determined inspection sequence, the starting point and the locations of all inspection points are connected sequentially to generate a continuous inspection path that can be executed by the inspection equipment.

[0070] The detailed steps for determining the order of inspection points are as follows.

[0071] Calculate the straight-line optical path data for two inspection points. For any two candidate inspection points, extract the coordinates of the two points and calculate the optical path data of the straight-line path between the two points in the layout model.

[0072] Point coordinates: The position value of each inspection point in the equipment coordinate system of the inspection equipment itself, which facilitates the positioning and navigation of the inspection robot; Straight-line optical path data: The data of the path segment formed by connecting two inspection points along a straight line, used to verify whether the straight-line path is unobstructed.

[0073] The ray tracing algorithm is used to determine whether two points are adjacent inspection point pairs. The algorithm processes the straight-line optical path data and the layout model of the two points. If the straight-line optical path and the obstacles in the layout model do not intersect, it means the path between the two points is unobstructed, and these two points are marked as selectable adjacent inspection point pairs. If they intersect, it means the path is blocked and direct passage is not possible, and they are not marked as adjacent point pairs.

[0074] Ray tracing algorithm: a graphics algorithm used to simulate the path of a ray propagating along a straight line and to determine whether the ray will intersect with objects in the scene. In this scheme, it is used to determine whether there are obstacles blocking the straight path between two points. Adjacent inspection point pairs: Two inspection points are connected by a smooth path and can be reached directly, so they can be arranged as two consecutive points in the inspection sequence.

[0075] The final inspection order is determined by a combination of security verification and distance constraints. First, the security verification agent assesses the security of all candidate adjacent inspection point pairs to confirm whether the path meets the security requirements. Then, the path planning agent, starting from the starting point, combines the distance from the starting point to each point, the security constraints of each point itself, and the results of the security verification to finally sort all the inspection points into the correct order.

[0076] Safety verification agent: A specialized agent in the crane agent cluster responsible for verifying the safety of the path plan.

[0077] The system receives task configuration information input by the user. It receives the inspection task configuration parameters input by the user through the terminal, specifying which locations the user needs to inspect and what data quality is required.

[0078] Task configuration information: The parameters entered by the user according to the inspection requirements include the resolution and clarity requirements of the acquired images (image acquisition conditions) and the location requirements of the components to be inspected (image location conditions).

[0079] The task parsing agent processes the data to obtain the inspection task. The task parsing agent performs natural language semantic understanding on the configuration information input by the user, and verifies whether the input information has logical contradictions or missing parameters. If the verification is successful, a clear and executable inspection task is generated.

[0080] Task parsing agent: A specialized agent in the crane agent cluster responsible for parsing user-inputted task information and verifying the rationality of the task, working based on the semantic understanding capabilities of a large model.

[0081] The component data includes collected multi-source data and corresponding collection locations. During verification, the data verification agent makes two judgments: first, whether the collection location meets the inspection location required by the task; second, whether the type and quality of the multi-source data meet the data conditions required by the task. If both judgments are satisfied, the component data passes the verification and meets the task requirements; if either judgment is not satisfied, it is judged as failing and needs to be collected again.

[0082] Task conditions: These are divided into inspection location conditions (the location where the data is collected must be consistent with the inspection points required by the task) and data conditions (the type and quality of the collected data must meet the task requirements, such as meeting the resolution requirements and ensuring that the data is complete and without missing data). Multi-source data: Multiple types of detection data collected from the same inspection point; Data verification agent: A specialized agent in the crane agent cluster responsible for verifying whether the collected component data meets the task requirements.

[0083] The feature extraction agent extracts features from all multi-source data collected at the same inspection point at the same time, and obtains the feature vector corresponding to each data.

[0084] Multi-source data: The multi-source data in this solution specifically includes: image data of crane components in different spectral bands (such as visible light images and near-infrared images), plus temperature data (component temperature obtained by infrared thermometry), ultrasonic data (internal defect data obtained by ultrasonic flaw detection), and point cloud data (three-dimensional contour data of components obtained by lidar scanning). Feature extraction agent: An agent in the cluster that is specifically responsible for extracting features that reflect the state of components from various multi-source data; Feature vector: Converting extracted features into digital vectors that can be processed by computers; Fusion agent: An agent in the cluster that is specifically responsible for fusing feature vectors of different types of data into a comprehensive feature; Multidimensional feature vector: A comprehensive feature vector containing multidimensional state information obtained by fusing features from different sources; Knowledge Network: A graph-structured knowledge network constructed with each component of the crane as a node and the connection relationships between components, historical fault records of components, and historical maintenance records of components as edges, to assist in condition diagnosis; Status diagnostic agent: An agent in the cluster specifically responsible for judging the status of devices based on fused features and knowledge networks; Graph Neural Networks: Deep learning neural networks specifically designed for processing graph-structured data. They are suitable for processing the knowledge network graph structure in this solution and can perform diagnostics by combining node features and graph structure relationships. Cluster consensus mechanism: Multiple agents in the cluster jointly verify the state diagnosis results. Only when the majority of agents confirm that the results are consistent will the final result be output, thus avoiding errors in judgment by a single agent.

[0085] Feature fusion: The fusion agent fuses all individual feature vectors to obtain a multidimensional feature vector V containing all feature information; Constructing a knowledge network: Using all equipment components of the target crane as nodes, and the connection relationships between components, the historical fault records and maintenance records of each component as edges, a knowledge network graph reflecting the component relationships and historical information is constructed. Diagnosis and confirmation: The state diagnosis agent is input into the trained graph neural network, which includes multidimensional feature vectors and knowledge networks, to obtain preliminary state diagnosis results. These results are then verified and confirmed through a cluster consensus mechanism, and finally, the determined device state information is output. Acquiring historical data: For each inspection point, the predictive agent retrieves all historical component data for that point within a set historical time window from the database.

[0086] Predictive agent: A specialized agent in the cluster responsible for predicting the remaining lifespan and failure probability of components based on historical data; Historical time window: A pre-defined historical time range, such as all historical detection data of this location within the past 12 months.

[0087] Analyze the changing trends: Perform time series analysis on multiple historical data to obtain the changing trends of the performance parameters of the corresponding equipment components at this point over time; Predict and incorporate inspection results: Based on changing trends, predict one or more of the following: remaining service life of equipment components, probability of failure in the next year, and performance degradation trend.

[0088] This application provides an artificial intelligence-based crane inspection method. This method introduces a cluster of crane intelligent agents created based on a large model, achieving intelligent verification and closed-loop control throughout the entire process, including inspection task configuration, path generation, data acquisition and verification, multi-source data collaborative processing, and equipment status diagnosis. The intelligent agent cluster can perform dual verification of inspection tasks and collected data, ensuring the safety and validity of the inspection process. Combined with the deployment model, it automatically generates optimized inspection paths and utilizes graph neural networks to fuse multi-source features and knowledge networks, significantly improving the accuracy and robustness of equipment status identification. Simultaneously, it predicts the remaining lifespan and failure probability of components through historical data analysis and maps structured inspection results to a virtual model in real time, generating a visual report, greatly improving the intelligence level, operational efficiency, and decision support capabilities of crane inspection.

[0089] Figure 14 An artificial intelligence-based crane inspection device is provided as an embodiment of this application. The device includes: The acquisition module 1410 is used to acquire the inspection tasks of at least one equipment component of the target crane and input each inspection task into the crane intelligent agent cluster created based on the large model. The verification module 1420 is used to verify the inspection task by the crane intelligent agent cluster. After the verification is successful, the deployment model corresponding to the target crane is called to generate an inspection path, triggering the inspection equipment to collect component data at each inspection point according to the inspection path, and the crane intelligent agent cluster verifies the collected component data. The inspection path includes the inspection points corresponding to each inspection task; the deployment model is a virtual model of the target crane. The collaboration module 1430 is used to determine the equipment status information of the corresponding inspection point by having the crane intelligent agent cluster collaboratively process the component data when the component data meets the task conditions of the inspection task and passes the verification of the crane intelligent agent cluster. The mapping module 1440 is used to generate structured inspection results based on the equipment status information, map the inspection results to the corresponding inspection points in the deployment model, and summarize the inspection results of multiple inspection points to generate a visual inspection report and push it to the user terminal.

[0090] The functions of each unit of the crane inspection device based on artificial intelligence provided in the above embodiments of this application can be realized through the above methods and steps. Therefore, the specific working process and beneficial effects of each unit in the crane inspection device based on artificial intelligence provided in the embodiments of this application will not be repeated here.

[0091] This application also provides an electronic device, such as... Figure 15As shown, it includes a processor 1510, a communication interface 1520, a memory 1530, and a communication bus 1540, wherein the processor 1510, the communication interface 1520, and the memory 1530 communicate with each other through the communication bus 1540.

[0092] Memory 1530 is used to store computer programs; When processor 1510 executes a program stored in memory 1530, it performs the following steps: Obtain the inspection task of at least one equipment component of the target crane and input each inspection task into a crane agent cluster created based on a large model; The crane intelligent agent cluster verifies the inspection task. After successful verification, it calls the deployment model corresponding to the target crane to generate an inspection path, triggering the inspection equipment to collect component data at each inspection point according to the inspection path. The crane intelligent agent cluster then verifies the collected component data. The inspection path includes the inspection points corresponding to each inspection task. The deployment model is a virtual model of the target crane. When the component data meets the task conditions of the inspection task and passes the verification of the crane intelligent agent cluster, the crane intelligent agent cluster performs collaborative processing on the component data to determine the equipment status information of the corresponding inspection point. Based on the equipment status information, a structured inspection result is generated, the inspection result is mapped to the location of the corresponding inspection point in the deployment model, and the inspection results of multiple inspection points are summarized to generate a visual inspection report and pushed to the user terminal.

[0093] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0094] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0095] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0096] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0097] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 13 The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.

[0098] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform an artificial intelligence-based crane inspection method as described in any of the above embodiments.

[0099] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the artificial intelligence-based crane inspection methods described in the above embodiments.

[0100] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments in this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0101] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0102] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0103] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0104] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected," "coupled," or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0105] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the embodiments in this application are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments in this application.

[0106] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the embodiments of this application and their equivalents, then these modifications and variations are also intended to be included in the embodiments of this application.

Claims

1. A crane inspection method based on artificial intelligence, characterized in that, The method includes: Obtain the inspection task of at least one equipment component of the target crane and input each inspection task into a crane agent cluster created based on a large model; The crane intelligent agent cluster verifies the inspection task. After successful verification, it calls the deployment model corresponding to the target crane to generate an inspection path, triggering the inspection equipment to collect component data at each inspection point according to the inspection path. The crane intelligent agent cluster then verifies the collected component data. The inspection path includes the inspection points corresponding to each inspection task. The deployment model is a virtual model of the target crane. When the component data meets the task conditions of the inspection task and passes the verification of the crane intelligent agent cluster, the crane intelligent agent cluster performs collaborative processing on the component data to determine the equipment status information of the corresponding inspection point. Based on the equipment status information, a structured inspection result is generated, the inspection result is mapped to the location of the corresponding inspection point in the deployment model, and the inspection results of multiple inspection points are summarized to generate a visual inspection report and pushed to the user terminal.

2. The method according to claim 1, characterized in that, Generate inspection paths, including: The path planning agent in the crane agent cluster determines the equipment location of the inspection equipment as the starting point of the inspection path; Based on the location of each inspection point in the deployment model and the starting point, and combined with the verification results of each agent in the crane agent cluster on obstacle information, safety constraints and historical path efficiency, the inspection sequence of each inspection point is determined. The inspection path is determined based on the inspection sequence and the location of each inspection point.

3. The method according to claim 2, characterized in that, Based on the location of each inspection point in the deployment model and the location of the equipment, the inspection sequence of each inspection point is determined, including: The coordinates of any two inspection points are processed to determine the straight-line optical path data of the two inspection points in the deployment model; the coordinates of the inspection points are the coordinates of the inspection points in the equipment coordinate system of the inspection equipment. A ray tracing algorithm is used to process the straight-line optical path data and the layout model. If the straight-line optical path data does not intersect with the obstacles in the layout model, then the corresponding two inspection points are determined as adjacent inspection point pairs. The safety verification agent in the crane agent cluster performs a safety assessment on each candidate adjacent inspection point pair, and the path planning agent determines the inspection sequence of each inspection point based on the distance between the starting point of the inspection path and each inspection point in the inspection point pair, the constraints of each inspection point, and the safety assessment results.

4. The method according to claim 1, characterized in that, The process of determining the inspection task includes: Receive task configuration information input by the user, the task configuration information including image acquisition conditions and image location conditions; The task parsing agent in the crane agent cluster performs semantic understanding and consistency verification on the task configuration information to determine the inspection task.

5. The method according to claim 1, characterized in that, The task conditions include inspection location conditions and data conditions; The component data includes multi-source data to be processed and the corresponding acquisition locations; The crane intelligent agent cluster performs real-time verification of the collected component data, including: the data verification intelligent agent in the crane intelligent agent cluster determines whether the collection location of the multi-source data meets the inspection location conditions, and whether the multi-source data meets the data conditions. If both conditions are met, then the component data satisfies the task conditions of the inspection task and passes verification.

6. The method according to claim 5, characterized in that, The multi-source data includes image data of the target crane's components in different spectral bands, as well as temperature data, ultrasonic data, and point cloud data of the target crane's components. The crane intelligent agent cluster collaboratively processes the component data to determine the equipment status information at the corresponding inspection points, including: The feature extraction agent in the crane intelligent agent cluster extracts features from image data collected at the same time under different spectral bands, as well as temperature data, ultrasonic data and point cloud data of the target crane components, and determines the corresponding feature vectors. The fusion agent fuses the feature vectors to obtain a multidimensional feature vector; A knowledge network is constructed using each equipment component as a node and the connection relationships between each equipment component, historical fault records, and maintenance records as edges. The state diagnosis agent in the crane agent cluster uses a trained graph neural network to process the multidimensional feature vector and the knowledge network, and after confirmation by the cluster consensus mechanism, determines the equipment state information.

7. The method according to claim 1, characterized in that, After determining the device status information, the method further includes: For any inspection point, the predictive agent in the crane agent cluster obtains multiple historical component data of that inspection point within a historical time window. Analyze multiple historical component data to determine the changing trend of the equipment components corresponding to the inspection point; Based on the aforementioned trends, one or more of the following are predicted: the remaining service life of the equipment components, the probability of failure, and the performance degradation trend. The prediction results are then added to the structured inspection results after being verified by the cluster.

8. An inspection device for a crane based on artificial intelligence, characterized in that, The device includes: The acquisition module is used to acquire the inspection tasks of at least one equipment component of the target crane and input each inspection task into the crane agent cluster created based on the large model. The verification module is used to verify the inspection task by the crane intelligent agent cluster. After the verification is successful, the module calls the deployment model corresponding to the target crane to generate an inspection path, triggers the inspection equipment to collect component data at each inspection point according to the inspection path, and the crane intelligent agent cluster verifies the collected component data. The inspection path includes the inspection points corresponding to each inspection task. The deployment model is a virtual model of the target crane. The collaboration module is used to determine the equipment status information of the corresponding inspection point by having the crane intelligent agent cluster collaboratively process the component data when the component data meets the task conditions of the inspection task and passes the verification of the crane intelligent agent cluster. The mapping module is used to generate structured inspection results based on the equipment status information, map the inspection results to the corresponding inspection points in the deployment model, and summarize the inspection results of multiple inspection points to generate a visual inspection report and push it to the user terminal.

9. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method described in any one of claims 1-7.

Images